Top 50 Analysis Using Boardgamegeek Data

Imports and Helper Functions

The Top 50 Lists

Top 10

Of all the games in a top 50 list, the most important are the top 10. They are the best of the best, the cream of the crop. Therefore, we give them special precedence here by making a nice display of their cover art.

The Full Lists

Game Crossovers

In this section, we check how many games appeared on multiple individual's lists. Each listed grouping shows the games that only they had in common.

Game Statistics

Comparing individual rankings against BGG rankings

In the following section, we compare each individual's ranking for a game against its ranking on Boardgamegeek. Because significant outliers in Boardgamegeek ranking can adversely affect the charts, we ignore any that are above the rank 2000. Omitting these games should have only a small effect on the results.

Here we plot each individual's rankings against each game's Boardgamegeek ranking. The more closely the points resemble a straight line, the more the individual's rankings match up with those on Boardgamegeek. The charts include a best fit line and a highlighted confidence interval. A larger confidence interval indicates that the rankings do not closely match the relative ordering for the games' BGG rankings.

The confidence intervals shown should be taken with a grain of salt, because they are greatly affected by outliers, which were removed to make these charts readable.

Here we see a box and whisker chart, showing the median and quartile ranges, as well as any outliers.

Finally, we have a breakdown of how many games were in different ranges of BGG rankings. The total ranking distribution does include multiple instances of the same game if there is crossover. Because of this, these charts should be seen as "how many game rankings were in this range?", rather than "how many games were in this range?".

Game weights

In this section we look at Boardgamegeek's "weight" for the selected games. The values range from 1 to 5, with more complex games scoring higher.

In the violin plots below, we can see the mean and quartile ranges as in the box and whisker plot, as well as the distribution of the data around that weight, shown as the width of the plot.

Player counts

In this section we will investigate the best player counts for games. Each game can support a certain number of players, and users on Boardgamegeek can vote "Best", "Recommended", and "Not Recommended" for each player count. Because a game may have multiple "best" player counts, we only consider the player count with the most "Best" votes as being the best for the purposes of this analysis.

Release Year

In this section, we investigate how each person's game selections break down by release dates. With such a wide range of possible dates, each chart only shows games from 1990 to today (selected semi-arbitrarily), but the omitted years for each individual are printed out above the charts.

User Rating Counts

We can also evaluate each individual's list by their relative popularity by using the game's number of ratings on Boardgamegeek. Games with a low number of ratings will likely either be unpopular games or very new games, so we can use this metric to determine which individual has the most "underground" or "cult of the new" list.

Here we see a plot of the popularity distribution for each user.

Game Mechanics Statistics

On Boardgamegeek, every game is labeled with a set of mechanics, such as "card drafting" or "worker placement". In this section, we look at how each person's selections are distributed with respect to these mechanics.

Note that every game can have a large number of listed mechanics. For example, a simple game like No Thanks! has five associated mechanisms, and a complex game like "Twilight Imperium 4" lists twelve. Because of this disparity, this analysis becomes biased toward more complex games, as they will tend to have more mechanics. In addition, some of the mechanics are so generic as to describe just about anything in a board game. Taking the top n mechanics for a game was discussed, but it turns out that this is not a valid approach. As an example, the top two mechanics listed for Ticket To Ride are "Card Drafting" and "End Game Bonuses". These are mechanisms present in the game, but a better description would probably come from the 3rd, 4th, and 6th in the list, "Hand Management", "Network and Route Building", and "Set Collection". Therefore, all mechanics for a game are included in the analysis.

Below, we chart the 10 mechanics that appear the most overall, as well as the mechanic distribution for each individual. As above with the BGG ranking distributions, the total distribution does count games multiple times. Loosely, this means that the total chart should be interpreted as answering the question "how many game selections were made with this mechanic?"

The charts all have high counts for mechanics like "Hand Management" and "Dice Rolling", but this information is all but useless to us, because just about every game that includes cards or dice requires you to manage those cards or roll the dice.

Game Families Statistics

In this section, we analyze statistics about each game's "family". On Boardgamegeek, a family can refer to a high-level mechanism (e.g. 4X), certain types of components (e.g. miniatures), or many other groupings. Looking at the families overall will not give useful information, so each subfamily will be looked at individually below.

Kickstarter

The board game community has wholeheartedly embraced Kickstarter, for better or worse. The "Crowdfunding: Kickstarter" family notes a game that has been released on the platform. Here we ask the simple question, "Which person has the most Kickstarter games on their list?"

Component Family

The "Component:" family is used to categorize the physical construction of the game, and how they are used.

Game Category Statistics

On Boardgamegeek, a game can be be assigned a set of categories that define the game at a high-level. These categories may describe the overall theme of the game (e.g. Pandemic falls under the "Medical" category), or they may describe what kind of a game it is (e.g. The Resistance is in the "Party Game" category).

The counts in this section are by user inclusions, not by game. To reiterate what has been said already in earlier sections, the overall chart should be seen as answering the question "how many selected games fall under this category?"